JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 109, A09206, doi:10.1029/2004JA010385, 2004

Ion cyclotron and heavy ion effects on reconnection in a global magnetotail R. M. Winglee Department of Earth and Space Sciences, University of Washington, Seattle, Washington, USA

Received 12 January 2004; revised 21 May 2004; accepted 8 July 2004; published 22 September 2004.

[1] Finite ion cyclotron effects play a significant role in determining the dynamics of the neutral sheet. The demagnetization of the ions facilitates reconnection and produces an electric field perpendicular to the direction of the tail currents. This in-plane electric field drives field-aligned currents and an out-of-plane (or core) magnetic field in conjunction with the generation of flux ropes. In addition to these electromagnetic effects, it is shown that ion cyclotron effects lead to the preferential convection of plasma from the dawnside to the duskside. This convection is consistent with results from single- particle tracking but differs from ideal MHD treatment where the flow occurs symmetrically around the Earth. A physical manifestation of these asymmetric particle trajectories is the wrapping of the field-aligned current between the region 1 currents and the region 2 and/or region 0 currents. In addition, localized density enhancements and depletions are seen in the tail where the local heavy ion density can be substantially elevated over ionospheric conditions. Because of the local density variations, reconnection across the tail is inhomogeneous. Reconnection is initiated postmidnight and then sweeps across to the dawn and dusk flanks within a few minutes. Because of this spatial variation, the ejection plasmoid is actually U-shaped and the subsequent flux rope formation is highly skewed. INDEX TERMS: 2744 Magnetospheric Physics: Magnetotail; 2753 Magnetospheric Physics: Numerical modeling; 2788 Magnetospheric Physics: Storms and substorms; 2736 Magnetospheric Physics: Magnetosphere/ionosphere interactions; 2431 Ionosphere: Ionosphere/ magnetosphere interactions (2736); KEYWORDS: reconnection, multifluid, simulation, magnetosphere, magnetotail, flux ropes Citation: Winglee, R. M. (2004), Ion cyclotron and heavy ion effects on reconnection in a global magnetotail, J. Geophys. Res., 109, A09206, doi:10.1029/2004JA010385.

1. Introduction and Hughes, 1992]. Lepping et al. [1995, 1996] used a force-free model for flux ropes and showed that in both the [2] Magnetic reconnection plays a critical role in deter- middle and distant tail regions the average diameter was mining the morphology of the magnetosphere during sub- 10 Re and in several instances may be as small as 1–3 Re. storms and storms. Evidence of reconnection has been The origin of these flux ropes in the magnetotail have been growing since the initial observations of negative Bz in attributed to the currents that are generated when the ions the tail associated with substorm onset [e.g., Hones, 1976]. become unmagnetized, while the electrons remain magne- While initial models showed a well-defined O-type neutral tized in the reconnection region. The demagnetization of the line within the plasmoid [Hones, 1979], subsequent studies ions leads to the violation of the frozen-in theorem con- have shown that plasmoids can have an appreciable core ditions and provides the physical mechanism by which magnetic field [Sibeck et al., 1984; Elphic et al., 1986; reconnection can occur. This resultant magnetic diffusion Slavin et al., 1989; Sibeck, 1990; Moldwin and Hughes, region is thought to have been observed during a Wind 1991, 1992]. On occasion, the core magnetic field can be as perigee pass [Øieroset et al., 2001]. strong as, and in some cases exceed, the lobe magnetic field [4] Theoretical and modeling efforts have increasingly [Slavin et al., 1995]. The direction of the core magnetic focused on resolving the microstructure of the reconnection field is usually (34 out of 39 cases) in the direction of the By region and thereby quantifying the controlling influences of component of the interplanetary magnetic field (IMF) the system. Drake et al. [1994] have shown for idealized [Moldwin and Hughes, 1992]. geometries that reconnection does not occur in the smooth [3] The core field is also observed to be generally MHD manner but rather occurs through filamentation and unidirectional, although, on occasion, bipolar By signatures kinking of the current sheet. Winglee and Steinolfson are seen in conjunction with bipolar Bz signatures [Moldwin [1993], Steinolfson and Winglee [1993], Biskamp et al. [1995], Ma and Bhattacharjee [1996], Zhu and Winglee Copyright 2004 by the American Geophysical Union. [1996], Pritchett et al. [1996], Shay et al. [1998], and Birn 0148-0227/04/2004JA010385 et al. [2001] have demonstrated, using a variety of different

A09206 1of17 A09206 WINGLEE: ION CYCLOTRON AND HEAVY ION EFFECTS A09206 codes (including full particle, hybrid, and multifluid), that symmetrically around the Earth. If we are to better model the Hall term in the generalized Ohm’s law plays an the dynamics of the magnetosphere, then this discrepancy important role in controlling the reconnection rate in colli- between particle and fluid perspectives needs to be resolved. sionless plasmas. The importance of the Hall term is that it [9] In section 2 the formalism behind a fully self-consistent describes the electromagnetic response to the system as the multifluid treatment with ion cyclotron terms included in the ions become demagnetized owing to their large ion gyrora- electrodynamics as well as the plasma dynamics is described. dius relative to that of the electrons, which remain magne- Note that it is these latter ion cyclotron terms that produce tized. As this demagnetization occurs, the above models all differential convections between the different ion species. show a highly structured reconnection region with a strong While the global code does not have the capacity to resolve all out-of-plane (or core) magnetic field. While the different the structures seen in high-resolution hybrid codes, it is models can show a variety of different instabilities (includ- demonstrated that Hall and ion cyclotron effects that are ing firehose, ballooning, lower hybrid drift, and kink- incorporated in the global simulations are already sufficient tearing), they share the common feature of a highly to alter the dynamics of the global magnetosphere. + structured magnetic field mapping that goes well beyond [10] The presence of heavy ions such as O also adds a that seen in ideal MHD models. new inherent scale length to the reconnection region. This [5] While the above particle and hybrid models are very scale length is fully resolved within the global code. In informative in determining the local structure of the recon- section 3 we demonstrated that while the overall magnitude nection, the one real problem that has received very little of the cross-polar cap potential and field-aligned currents attention is how to place this physics into global models so remain of the order of that seen in previous simulations, the that direct comparisons with in situ observations of magne- full inclusion of the ion cyclotron terms modifies the tospheric activity can be made. Initial global simulations structure of the auroral current system to produce the spiral [Winglee et al., 1998] incorporating the Hall and rPinthe flow of current between the region 1 and region 2 currents Ohm’s law have shown similar core magnetic field and flux as noted by Iijima and Potemra [1976, 1978] using data ropes structures, as seen in the above particle and hybrid from the Triad satellite. In other words these apparently codes but on the observed scale size of a few Re. This initial small corrections have global consequences for magneto- work assumed the presence of only a single ions species spheric activity. In section 4 we then show the actual within the magnetosphere. structures generated in the tail including the dawn-dusk [6] However, the assumption of a single ion species for acceleration of plasma and asymmetric reconnection rate the magnetosphere is probably not accurate, since during across the tail, with a corresponding distortion of the disturbed times the ionosphere can be an important source plasmoid/flux rope topology. A summary of results is given of plasma to the magnetotail [Chappell et al., 1987] and this in section 5. outflow can be dominated by heavy ionospheric ions such + as O [e.g., Yau and Andre´, 1997]. To address this issue, 2. Simulation Model Winglee [2000] developed a global model that included the presence of heavy ions. This model as able to provide three- [11] In this section we detail the equations used in the dimensional (3-D) rendering of the geopause (i.e., the multifluid modeling, including for the first time ion cyclo- boundary within the magnetosphere where the plasma tron corrections to the momentum equations of the different changes from one being dominated by plasma of solar wind ions species present. A discussion is presented on the origin to one being dominated by plasma of ionospheric relative size of these corrections and why they can at least origin). More recently, these ionospheric outflows have partially be incorporated into a global model. We also been shown to produce saturation of the cross-polar cap compare and contrast the code capabilities relative to hybrid potential [Winglee et al., 2002]. codes to give more insight into the assumptions being made [7] In this paper we demonstrate that ion cyclotron effects within the code. The code details are then described. are not only important with respect to the electrodynamics 2.1. Electrodynamics of the system due to corrections in the generalized Ohm’s law, but they also have an important effect on the momen- [12] The dynamics of any fluid plasma component are tum equations for the different ions species. These effects given by are shown to play a critical role in determining the particle acceleration and convection in the reconnection region and @ra þrÁðÞ¼raV a 0 ð1Þ in the structure of the field-aligned currents into the auroral @t region. Such plasma flows are then shown to modify the reconnection rate across the magnetotail, which in turn dV GM r a ¼ q n ðÞrE þ V BrðÞ P J r ~r ð2Þ modifies the overall structure of flux ropes generated in a dt a a a a R2 a association with substorm onset. [8] This work is motivated by the fact that recent single- particle tracking results within the electric and magnetic @P a ¼grÁðÞþP V ðÞg 1 V ÁrP ; ð3Þ field from global simulations show dawn-dusk acceleration @t a a a a of plasma [Winglee, 2003] consistent with Speiser [1965] particle motion. However, this asymmetric particle acceler- where ra, Pa, and Va are the mass density, pressure, and ation does not appear within the ideal MHD. Instead, for velocity of the component a. MHD is based on combining southward interplanetary magnetic field (IMF) the flow of the above equations to give a single-fluid treatment. The plasma from the reconnection region is predicted to occur multifluid treatment is based on the same equations, but the

2of17 A09206 WINGLEE: ION CYCLOTRON AND HEAVY ION EFFECTS A09206 dynamics of the electrons and the different ion species are suppose that the magnetotail is driven into lobe reconnec- kept separate. For electrons it is assumed that they have tion (i.e., that field lines that were initially in the lobe have sufficiently high mobility along the field lines that they are sufficient time to convect into the plasma sheet and be approximately in steady-state drift motion. In other reconnected). The lobe field is of the order of 20 nT, and ne words their motion can be described by drift motion (i.e., is of the order 1cm3 or less. Suppose that the global me dVde/dt = 0) so that code can at best resolve the current sheet at a minimum of 3 grid units so that the inherent length scale is 3 Dx and (J 1 B/en)/(V B) is of order 20–40%, that is, the strength of E þ V de B þ rPe ¼ 0: ð4Þ ene the correction term is significant and cannot be neglected. [16] A similar calculation would show that these correc- Equation (4) is not actually solved for Vde but instead is tions are important at the magnetopause but are essentially used as an Ohm’s law to give E, where Vde is determined negligible in the inner magnetosphere where the densities from the induced currents and the assumption of quasi- are much higher. Since these reconnection regions deter- neutrality, i.e., mine the particle entry into the magnetosphere and play a critical part in the energization of plasma within the mag- X X ni J 1 netosphere, it is an important starting point to incorporate ne ¼ ni; V de ¼ V i ; J ¼ rB: ð5Þ ne ene m0 these nonideal MHD effects into a global code. Presently, i i there is no other means to incorporate these effects on Thus substitution of equation (5) into equation (4) yields the reconnection on global scale lengths. modified Ohm’s law of 2.2. Plasma Dynamics [17] The description of the electron dynamics is completed X ni J B 1 by the pressure equation E ¼ V i B þ rPe þ hðÞx J: ð6Þ i ne ene ene @Pe ¼grÁðÞþPeV de ðÞg 1 V de ÁrPe ð7Þ The first term in equation (6) is the ideal Ohm’s law and the @t last term h(x)J is added to allow for finite conductivity in the ionosphere. Collisions beyond this region are assumed and the evolution of the magnetic field by the induction to be negligible. No anomalous resistivity is included in the equation code, as the nonideal terms included in equation (6) are @B sufficient to drive reconnection. þrE ¼ 0: ð8Þ [13] The other terms in equation (6) include the Hall and @t rP terms and are ion cyclotron corrections to the Ohm’s law. Such corrections to the Ohm’s law are known to be The ion dynamics is determined by simultaneously time important in terrestrial reconnection studies and can drive stepping all the individual ion species equations (i.e., field-aligned currents into the auroral region [Zhu and equations (1)–(3)) using the electric field in equation (6). Winglee, 1997; Winglee et al., 1998; Shay et al., 1998]. These equations are essentially the same as used in hybrid These terms in dimensionless parameters are of order of the codes except that they are in the fluid limit as opposed to ion skin depth (c/w , where w is the ion plasma frequency) having particle ions. In section 2.3 we will discuss why pi pi these numerics can lead to stable solutions despite the fact relative to the grid spacing Dx. In the following, Dx is 0.25 Re (1600 km) in the vicinity of both the dayside and nightside that ion skin depth is only a fraction of the grid spacing. reconnection regions. The ion skin depth for protons Before discussing the numerics, though, it is important to 3 look at an explicit substitution of the ion momentum assuming a density of 1 cm is 200 km. So for the equation using equation (6), which has the form global model we can set Dx/(c/wpi)to8, whereas in hybrid models this ratio is more typically set to unity. If the plasma ! sheet density should decline, say during the growth phase of X dV a qana ni a substorm, then Dx/(c/wpi) can decrease to 2–4, so it is ra ¼ ðÞrJ B rPe Pa þ qana V a V i dt ene ne not appropriate to assume that this term is always negligible i as assumed in ideal MHD. GMJ B þ qanahJ ra~r: ð9Þ [14] One might argue that at a value of 8 the global code R2 cannot resolve any ion cyclotron effects in the electro- dynamics. For protons, even at a value of 8, gradient B Equation (9) is essentially the same that has been used in effects are nonnegligible. For oxygen ions, their gyroradius previous multifluid treatments [e.g., Winglee, 2003; Winglee becomes resolvable when they enter the tail and magneto- etP al., 2002] except that the velocity difference term (Va pause current sheets, particularly if they are accelerated to (ni/ne)Vi) is now retained. In MHD all the ion species are perpendicular velocities above 100 km/s. assumed to have the same drift velocity and hence the [15] Finite gyroradius effects are not the whole story, as neglecting of this term. This assumption makes a con- one also needs to take into account collective effects. venient starting point to evaluate point differences in the Consider the actual size of the V B terms relative to global dynamics owing to the presence of heavy ions, the J B/en term. While fast flows are seen in the in-plane including the saturation of the cross-polar cap potential. direction, the speeds of plasma flowing in the direction of [18] However, neglecting the above term does not provide the current sheet are of the order of 50–100 km/s. Now a complete description of the system. In fact the fluid flows

3of17 A09206 WINGLEE: ION CYCLOTRON AND HEAVY ION EFFECTS A09206 that develop with the dropping of this term produces the additional constraint. In particular it requires the simulation symmetric plasma flows around the inner magnetosphere time step to be reduced to resolve such waves. With 0.25 Re for southward IMF. This is contrary to the particle-tracking resolution and 2.5 Re inner radius, this requirement produ- results that show dawn-dusk acceleration of ions across the ces about a 25–50% reduction in time step, which is not too tail current sheet. This difference between the actual fluid much of a hardship on the code. velocity and a center of mass velocity is very similar to that [23] One of the outstanding issues in reconnection is how in particle treatments. In the latter, one splits the particle to produce the demagnetization of the electrons. This is motion in terms of a drift motion and cyclotron rotation. produced in the present code by the generation of field- The small difference in speed that arises from the particle aligned currents, which are incorporated by the inclusion of gyrating in an inhomogeneous magnetic field drives the rB the gradient pressure term in the Ohm’s law. As reconnec- drift, and this occurs even if the gyroradius is very much tion occurs, the field-aligned currents remove electrons from smaller than the scale length of the gradient in the magnetic the reconnection region and the return current places away field. The same is true in the present case in that if there are from the reconnection region to match the accelerated ion gradients in the magnetic field, then the velocity difference flows. Effects from the direct acceleration of electrons term is nonzero and thereby drives the breaking of the within the reconnection region are not included in the symmetric flows of ideal MHD. present model, as we are unable to resolve electron cyclo- [19] In the presence of different ion species the velocity tron effects. It is not known at this time whether the difference term is in general nonzero owing to finite ion relocation of the electrons from field-aligned currents or cyclotron corrections. For example, consider a typical their direct acceleration in the reconnection region is the current sheet configuration with a magnetic field of 5 nT dominant process in pulling electrons off the reconnection near the center of the current sheet. For a keV proton the field lines. However, the development of field-aligned gyroradius is 600 km. For an O+ ion at the same energy currents as described here has been seen in full particle the gyroradius is 2400 km, which starts to become resolved simulations, albeit in more restricted geometries [Winglee in the global simulations and can therefore start to experi- and Steinolfson, 1993; Zhu and Winglee, 1996]. ence electric field acceleration. A subtle but important point is that even if the grid resolution is not sufficient to resolve a 2.3. Numerical Algorithm and Boundary Conditions gyroradius, a particle or fluid element undergoing a gyration [24] The above equations are solved for a three-ion within a grid unit will still experience the gradient in the component plasma: solar wind protons, ionospheric protons, magnetic field (since all properties rely on interpolation and ionospheric O+ ions. On including the electron dynam- between grid points) so that even without full resolution rB ics, the code provides a four-fluid description of the global effects/drifts will still develop if the bulk temperature magnetosphere. The equations for each of these components becomes sufficiently high to become comparable to the is solved using a two-step Lax-Wendroff differencing convective drifts. scheme [Richtmyer and Morton, 1967] with Lapidus + [20] It is also important to remember that the O ions will smoothing on plasma properties only [Sod, 1978]. The latter not remain at the same energy as the protons but in fact will is required to remove unphysical grid point oscillations experience preferential acceleration and therefore have an across sharp discontinuities such as the bow shock. While even larger energy and gyroradius. Indeed the particle the two-step Lax-Wendroff scheme is one of the simpler trajectory analysis of Speiser [1965] shows that ions that numerical methods, its simplicity allows a tractable means are ejected from the reconnection region will have speeds of for including much more complicated but more realistic the order of the Alfve´n speed irrespective of mass so that the physics. Another reason for using this scheme is that we can heavy ions will always tend to be preferentially accelerated track how the evolution of the system changes as additional over light ions. It is for the same reason that protons will be ion cyclotron terms are added relative to previously pub- preferentially accelerated over electrons in reconnection lished treatments [Winglee et al., 1998, 2002; Winglee, regions. 2000]. As shown in the following section, because the [21] One could suppose that maybe the heavy ions have scheme is well validated the development of dawn-dusk only a low concentration and possibly have little effect. asymmetries in the magnetosphere is due to physical However, the results of Seki et al. [1998, 2001] have shown processes associated with the inclusion of ion cyclotron that O+ flows can be typically a few percent in the tail. At and heavy ion mass loading effects and is not due to just 6% number density, the O+ ions will provide equivalent numerics. mass as the protons, and if they attain comparable veloci- [25] At this point there is a major divergence between the ties, as mentioned above, they will also be responsible for multifluid codes and the hybrid codes. Hybrid codes accu- equal amounts of energy transport out of the current sheet. mulate the ion density on the grid to be used in the Ohm’s Thus the plasma dynamics generated by this seemingly law given by equation (6). As such they have the advantage small term can have an appreciable effect on the overall of being able to incorporate effects from non-Maxwellian global properties of the magnetosphere. This will also be distributions, which the fluid code cannot. Hybrid codes demonstrated by the simulation results in the following have the disadvantage, though, in that they are subject to sections. grid point noise due to particle statistics. Such grid point [22] One of the consequences of retaining this term is that noise, particularly in the electric field, can eventually lead to it generates ion cyclotron waves in the system. These ion numerical instabilities. The use of Lapidus smoothing in the cyclotron waves appear at both the proton and heavy ion fluid code eliminates the growth of such grid point insta- cyclotron frequencies. The heavy ion frequency is not a bilities. It is for this reason that the multifluid code can problem, but the proton cyclotron frequency does add an operate in a regime where the ion skin depth is only a

4of17 A09206 WINGLEE: ION CYCLOTRON AND HEAVY ION EFFECTS A09206 fraction of a grid spacing. The downside, of course, is that tating energetic plasma is lost from the system. At the outer we lose all the physics associated with very small scale boundary the flows are supersonic so that open boundary processes seen in higher-resolution hybrid code, but we conditions are used, except at the left-hand boundary where have the advantage that a full global magnetospheric model the solar wind conditions are specified. can be developed with ion cyclotron effects incorporated in [29] The temperature of the plasma is kept low over the key regions. Here we show that even with this limited poles but is increased toward the equator to take into treatment such ion cyclotron terms can produce substantial account the presence of hotter trapped populations there. effects. The equations are solved on a structured Cartesian To ensure equilibrium along each flux tube, they are initially grid using a box-in-box grid system. loaded with a constant value (set by mapping the simulation [26] The grid system allows for high 0.25 Re resolution grid point along the dipole field line to the equator and over much of the inner plasma sheet, while allowing the giving all points along the field line the value prescribed to incorporation of a total system size of 200 Re down tail and the equator). The equatorial bulk temperature is set to ±50 Re in the y, z directions at 1–2 Re resolution. The solar 60 eV and in the polar cap it is less than 0.3 eV. The wind boundary is at x =35Re. The inner boundary is set at low temperature over the polar cap is typical of conditions + 2.5 Re. To simulate the Earth’s resistivity, the region within there and ensures that much of the ionospheric O is the inner boundary is given a finite resistance equivalent to gravitationally bound, and forcing by the solar wind (for a Reynolds number of 10. At the actual inner boundary example, by centrifugal acceleration [Cladis, 1986]) is (representing the ionosphere), the Reynolds number is required to drive heavy ion outflows. increased to 20 and at one grid point above it is set at 40. [30] An equilibrium for the magnetosphere is established At all other points the resistivity is zero. Zero dipole tilt is by blowing in the solar wind from the left-hand boundary assumed in the following. for a period of 2.75 hours. The solar wind density is held 3 [27] The choice of placing the resistivity at the inner fixed at 5 cm and a speed of 450 km/s with zero IMF. boundary provides only a crude model for the ionosphere. Having established this equilibrium, a southward IMF of We have performed sensitivity tests of the results, and to 8 nT is imposed for 80 min. This time scale ensures date key indicators such as the cross-polar cap potential dissipation of transients associated with the southward have been shown to be insensitive to the choice of the turning of the IMF and the ejection of a tail plasmoid. A ionospheric resistivity. The reason for this insensitivity is northward turning of the IMF of +8 nT is then imposed to that the ionospheric response is not just due to its conduc- investigate the release of energy stored in the current sheet. tivity but also through the ionospheric outflow rate. The reason for this insensitivity can be thought as follows. The amount of solar wind momentum transferred to the magne- 3. Ionospheric Response tosphere is, to first order, determined by the size of the [31] In this section we establish the fact that the ion magnetosphere and its ability to couple to it, which is cyclotron effects do not change the overall response of the controlled by prevailing IMF conditions and the solar wind system in the form of the total auroral currents and cross- speed and density. This momentum transfer therefore sets polar cap potential, but they do change the structure/spatial the speeds of the convective flows within the magneto- distribution of the auroral currents. To see this, Figure 1 sphere. These flows set the convection electric field through shows the timing of the imposed IMF changes relative to some type of Ohm’s law such as given in equation (6). the induced changes in the integrated field-aligned currents Integration of this electric field sets the cross-polar cap into the northern hemisphere (positive is for downward potential. Thus changing the conductivity will have little currents and negative is for upward currents) and the effect on the momentum transfer and therefore little effect corresponding cross-polar cap potential. It is seen that the on the cross-polar cap potential. southward IMF can induce currents of the order of a few [28] However, if the mass loading of the magnetosphere MA, which is 10 times greater than the value for north- is altered by induced ionospheric outflows, there will be a ward IMF conditions. The cross-polar potential is driven corresponding change in cross-polar cap potential which in from 20 kV to a steady-state value of 50 kV. These turn will appear as a change in the overall conductivity of values are comparable to other substorm case studies the system (for example if the potential drops without a [Weimer et al., 1992]. corresponding drop in field-aligned current). This effect has [32] An important feature to note in Figure 1 is that the been demonstrated by Winglee et al. [2002]. Thus the key buildup of the auroral currents is slower than the buildup of point here is that the ionospheric response is just not the cross-polar cap potential. The latter reaches a peak value conductivity but must also include the known ionospheric at 70 kV by T = 0300 and then subsides to its steady-state outflows mentioned in the introduction. To this end, we value, whereas the field-aligned currents have a period of place ionospheric protons on the inner boundary with a rapid increase up to T = 0308 and then maintain a steady density of 1000 cm3 and we have a fixed O+ concentration increase up to the time when the northward IMF reaches the of 5%. We have retained only a small O+ concentration in magnetosphere. this paper to demonstrate that even for very small concen- [33] The data in Figure 1 can be used to attain useful trations of O+ the tail dynamics can be substantially characteristics of the ionospheric response in terms of V/I modified over that from just a single fluid treatment such for an effective ionospheric resistivity (Figure 2a) and V I asMHD.Asdiscussedinthefollowingsection,these for an estimate of the Ohmic power into the ionosphere boundary conditions produce outflow rates of the order of (Figure 2b). It is seen from these figures that while the 1025 H+ ions/s and 1024 O+ ions/s. Because the density and actual ionospheric resistivity is held fixed, the effective temperature at the inner boundary are held fixed, precipi- resistivity of the ionosphere drops during the southward

5of17 A09206 WINGLEE: ION CYCLOTRON AND HEAVY ION EFFECTS A09206

Figure 1. Time history of (a) the integrated field-aligned currents (IA) into the ionosphere (black for downward currents and gray for upward currents) and (b) the cross polar cap potential (VP). A 2.75 hour period of zero IMF is used to produce an initial equilibrium magnetosphere. The focus is then on the response of the magnetosphere to a subsequent 80 min period of strong southward (8 nT) IMF followed by 60 min of strong northward IMF. The cross polar cap potential responses much faster than the currents, similar to a highly inductive system.

IMF period by a factor of 4. The magnitude of the iono- the light ion outflow. This relative increase arises because spheric resistivity and the percentage decrease are consistent the heavy ions are being energized to overcome gravity at with observed characteristics of ionospheric resistivity. lower altitudes to create outflows substantially larger than The magnitude of the Ohmic power shown in Figure 2b would have occurred from their thermal outflow alone. of 100 GW is also consistent with typical estimates for [35] It is important to note that the magnitude of the light power dissipation in a substorm [e.g., Lu et al., 1996]. Note ion outflow is consistent with the results of Yau and Andre´ that while the cross-polar cap potential reaches a steady- [1997], but if anything we are underestimating the heavy state value, the continuous rise in the auroral currents during ionospheric outflow by at least a factor of 2. This under- the period of southward IMF leads to increasing power estimation arises because the ionospheric O+ relative dissipation in the ionosphere until the effects of the north- density is held fixed at 5%. Enhancing ionospheric outflows ward turning in the IMF appear in the ionosphere. would produce further modification of the cross-polar cap [34] The response of the cross-polar cap potential can be potential [Winglee et al., 2002]. The important point is that understood from the induced ionospheric outflows. These as far as the overall ionospheric response is concerned, the outflows are shown in Figure 3 as they pass through a boundary conditions assumed in the code (as described in sphere located at 4 Re. Prior to the arrival of the southward section 2.3) are not extreme and appear to quantitatively IMF, the O+ outflow is 1/50 of the ionospheric H+ reproduce typical ionospheric bulk properties. outflow. This ratio is smaller than the assumed relative [36] The difference in the time response of the heavy and concentration at the ionosphere because with little forcing light ionospheric ions is further illustrated in Figure 3b, the heavy ion outflow is gravitationally bound. With the which replots the outflows in terms of the ionospheric arrival of the southward IMF, the ion outflow for both currents calculated from the simulations. It is seen that both species increases but with the light ion outflow reaching the light and heavy outflows show a hysteresis-like curve, saturation first at T 0300. The heavy ions with their larger but the response time is different for the light and heavy inertia reach their peak value much later at T = 0330. At this ions. For the light ions the curve associated with increasing point the relative outflow of heavy ions is 1/6 to 1/10 of current (southward IMF) essentially overlaps the curve for

6of17 A09206 WINGLEE: ION CYCLOTRON AND HEAVY ION EFFECTS A09206

decreasing current (northward IMF). For the heavy ions the rate of increase with auroral current is steeper than for the light ions and hangs longer at high values during the initially northward turning of the IMF but eventually returns to its initial quiet time value. The reason for this longer hang time is that the outflow is measured at 4 Re, so it takes some time for the heated heavy ionospheric ions to reach the measuring point. At the end of the period considered, the O+ flux is seen to be dropping very much faster than the H+ outflow. [37] The effect of ion cyclotron terms in the actual structure of the aurora current system is shown in Figure 4. For north/south IMF conditions the mapping of the field-aligned currents in ideal MHD should be antisym- metric about the noon-midnight meridian. At T = 0250 this is approximately the case, although for zero IMF there is a slight asymmetry with upward current region extending slightly farther down in latitude that the downward current region. As the IMF turns southward (T = 0313), there is both an increase in the peak value of the current density and an expansion in latitude of the current systems. Beyond this time, the peak value does not change significantly but the area does (e.g., T = 0337 and 0410). This increasing area corresponds to the period of slow increase in the total current seen in Figure 1. Note that as the area is increasing the asymmetry in the current system also increases, with the dawnside downward region 1 current system wrapping around through noon to the dusk region 2 current, while the duskside upward region 1 wraps through midnight onto the dawnside region 2 current system. This nightside

Figure 2. (a) The effective plasma resistivity (VP/IA) and wrapping of the current extends into the period of northward (b) the effective Ohmic power (VP IA) using the data in IMF for 20 min but eventually fades as the magnetosphere Figure 1. The effective plasma resistivity drops by nearly a becomes dipole-like under northward IMF. During the quarter during the period of southward IMF, whereas the northward IMF the dayside wrapping of the near-noon Ohmic power shows an initial steep rise to 80–100 GW region 1 current into the region 0 current becomes more associated with the rapid raise in the cross polar cap pronounced (e.g., T = 0420 and 0430). potential, followed by a slower increase to a peak power of [38] The wrapping of the auroral current systems seen in 130 GW as the auroral currents slowly increase under Figure 4 is an important feature. The wrapping of the continued southward IMF. current systems was first noted by Iijima and Potemra

Figure 3. The induced ionospheric outflows (calculated at a radius of 4 Re) for the assumed ionospheric conditions as a function of (a) time and (b) the auroral currents as shown in Figure 1. The magnitudes of the outflows are comparable to fluxes report by Yau and Andre [1997] with the O+ flux if anything being on the low side. As a function of auroral current, the outflows form hysteresis curves with the heavy ions having a sharper slope and a longer recovery time.

7of17 A09206 WINGLEE: ION CYCLOTRON AND HEAVY ION EFFECTS A09206

Figure 4. Evolution of the spatial distribution of the auroral currents. Upward current regions are shown in green and the downward currents are shown in yellow. The retention of ion cyclotron terms in momentum equation breaks the symmetry in the auroral current system to produce wrapping of the nightside dusk region 1 current to the dawnside region 2 currents and the dayside dusk region 1 current to the region 0 dawn current and the dayside dawn region 1 current to the dusk region 2 current. The latter two connections are seen to continue after the northward turning of the IMF.

[1976, 1978], using statistical maps of the field-aligned [1976, 1978] pattern. The main discrepancy is that our currents. Their results are shown in Figure 5a. One can region 2 current is not intense enough to show at local check the physical origin of this feature evaluating the times near midnight. This probably arises because much of auroral currents with different ion cyclotron corrections the ring current is carried by particles in the tail of the included. Figure 5b shows the results under exactly the distribution, and the physics of such energetic tails is not same conditions but only the ideal MHD equations are well incorporated in fluid treatments. solved. These results, similar to Ogino et al. [1986], show an antisymmetric current system and no wrapping of the aurora current across the noon-midnight meridian is present. 4. Magnetospheric Response Figure 5c shows the results using the original version of the 4.1. Density Variations multifluid simulations [e.g., Winglee, 2000; Winglee et al., [40] In this section we consider the changes in the 2002]. This version includes the Hall and gradient pressure structure of the magnetosphere associated with the field- corrections to the Ohm’s law but neglects the finite ion aligned currents and ionospheric outflows described in the cyclotron terms in the momentum equation (i.e., all the ions previous section. The overall behavior of the composition of were assumed to have the same perpendicular drift speed). the magnetosphere is similar to previous multifluid simu- The same O+ concentration of 5% is assumed. It is seen lations [Winglee, 2000] in that with the turning of the IMF in this case that there is partial breaking of the MHD to increasingly stronger southward IMF, the solar wind antisymmetry with partial extension of the dusk region 1 density within the magnetosphere drops. This is illustrated current across the noon-midnight meridian but nothing to in Figures 6a and 6b where the lobes first become depleted the extent reported by Iijima and Potemra [1976, 1978]. and the current sheet thins (as seen by the cross-plane cuts). [39] Figure 5d shows the present results with ion cyclo- However, a key difference is that this thinning does not tron corrections in both the Ohm’s law and momentum occur at a uniform rate across the tail, and instead there is a equations. It is seen that the wrapping of the nightside thick high-latitude band of solar wind plasma that penetrates currents is enhanced with the dawnside and duskside into the lobe across the duskside. region 1 currents crossing the noon-midnight meridian both [41] After 30 min from the southward IMF turning, the near noon and near midnight. The present system produces contributions of the solar wind plasma to the tail current the best agreement to date with the Iijima and Potemra sheet falls to lobe-like densities at less than 0.01 cm3

8of17 A09206 WINGLEE: ION CYCLOTRON AND HEAVY ION EFFECTS A09206

Figure 5. Comparison of the auroral field-aligned current systems from (a) Iijima and Potemra [1978], (b) ideal MHD, (c) multifluid treatment with ion cyclotron corrections in the Ohm’s law only, and (d) the present multifluid treatment with ion cyclotron treatments in the Ohm’s law and momentum equations. As the nonideal effects are added, the simulation results converged on the wrapped field-aligned currents of Iijima and Potemra [1978].

(Figures 6c and 6d). The last region of the current sheet to plasma sheet increase (Figure 7c), except near the deple- be depleted of solar wind plasma is on the premidnight tions regions, which actually increase in size. The depletion sector. This lag occurs because the plasma is subject to regions develop because the nightside source is convected current sheet acceleration that preferentially transports the into the sheet closer in toward the Earth, while the dayside plasma from the dawnside to the duskside. With the (cusp/cleft) source convects much deeper into the tail, northward turning of the IMF, solar wind plasma enters similar to the particle tracking results of Winglee [2003]. through the flanks (by high-latitude reconnections and the [43] As the current sheet continues to thin under the subsequent convection of newly closed field lines around influence of southward IMF, the current sheet tends to be + the flanks). The midtail region at 30–60 Re is first to refill relatively depleted of ionospheric H , similar to the solar (Figure 6f), with the dawnside having greater penetration of wind protons. There is a limited range of local times where solar wind plasma than the duskside. high-density channels appear, and these channels, as shown [42] The dawn-dusk asymmetries are even more apparent below, are associated with reconnection in the tail. These in the density of the ionospheric contributions to the channels lie preferentially on the duskside (Figure 7d) and magnetosphere, as shown in Figures 7 and 8. At the first are seen only if the ion cyclotron term is kept in the time shown in Figure 7, the ionospheric H+ ions are present momentum equation. Another asymmetry that develops is through most of the current sheet except for the depletion a dayside bulge associated with the enhanced plasmaspheric regions near the inner edge of the plasma sheet. Because of plasma convection near noon and is associated with wrap- the relatively thick current sheet (since the forcing from the ping of the region 1 and 2 currents shown in Figure 4. + IMF is relatively weak at this time), ionospheric H is seen [44] The evolution of the heavy ion density shows stron- at high latitudes as well as in the cross-tail cuts in Figure 7a. ger ion cyclotron effects, but because their gyroradius tends With strong southward IMF, this high-latitude plasma is to be much larger than the proton gyroradius these features convected into the plasma sheet so that while their density do not necessarily coincide with those of the ionospheric in the lobe decreases in Figure 7b, their contributions to the protons. This is seen for example in Figure 8a where a

9of17 A09206 WINGLEE: ION CYCLOTRON AND HEAVY ION EFFECTS A09206

Figure 6. Evolution of the solar wind density as seen by equatorial density contours and three cross-tail cuts at x = 15, 30, and 60 Re. Southward IMF leads to the depletion of the solar wind component within the magnetosphere with the duskside being the last to become fully depleted. The northward turning of the IMF leads to the reentry of solar wind plasma by high-altitude reconnection and subsequent convection of the newly reconnected field lines into the magnetosphere.

Figure 7. As in Figure 6, except the evolution of the density of ionospheric H+ is shown. The H+ density in the magnetosphere also declines during southward IMF, but there are narrow filaments of enhanced density that extend down the tail between midnight and the dusk magnetopause. These filaments are only seen when ion cyclotron effects are included in the ion momentum equation.

10 of 17 A09206 WINGLEE: ION CYCLOTRON AND HEAVY ION EFFECTS A09206

Figure 8. As in Figure 6 except for ionospheric O+ ions. The dawnside under southward IMF is initially loaded with the heavy ions, similar to single particle trajectories of Winglee [2003]. This plasma is convected across to the duskside to produce a bulge in the plasma density near the dusk terminator. An additional bulge near noon is also seen. dawn-dusk asymmetry at early times during zero IMF is that is rich in O+, particularly on the duskside. With the much more evident. The southward IMF enables convection northward turning of the IMF the O+ rich regions in of the heavy ions into the near-Earth plasma sheet, but the the equatorial regions fade, leaving only local regions in thin channels seen in the protons in Figures 6c and 6d are the lobe with a high O+ concentrations. not present in the O+ in Figures 8c and 8d. The loading of the plasma sheet is seen to spread from postmidnight across 4.2. Effects on Reconnection to the dusk terminator (Figures 8b–8e). Thus similar to the [47] Because of the inhomogeneities in the plasma den- single particle trajectories of Winglee [2003], heavy ion sities, the reconnection rate across the tail varies between outflows from the dawn to midnight sectors feed the tail the dawn and dusk flanks. The reason for this is that the current sheet, and this plasma is accelerated across to the convection of plasma into the reconnection region depends dusk sector. P on the Alfve´n speed, and this is a function of the plasma + [45] The relative density of O (i.e., (nO/ ni)) is shown density, which in turn is dependent on the plasma compo- in Figure 9 and highlights the differences associated sition. If plasma is convected from the dawnside to the with the different convection patterns of the different ions duskside as seen in the previous subsection, then the species. Initially (Figure 9a), the O+ concentration in the reconnection rate will occur relatively faster on the dawn- inner magnetosphere is at the 5% level or lower except in side than on the duskside. To see these differences in the + the lobe, where the fast speed of the light ions leaves an O reconnection rate, Figure 10 shows the evolution of Bz in rich region at the high latitudes in the lobes. As the IMF the equatorial region leading up to reconnection in the tail. turns southward, the greater inertia of the heavy ions causes The figure uses a saturated color table to highlight the them to be the dominant species in the plasma sheet separation of regions of positive Bz (red) from negative Bz boundary layer (PSBL), as seen in Figure 9b. At this stage, (blue). The exact position of the neutral line (Bz =0)is despite the fact that the ionospheric O+ relative concentra- shown as the black curve in the plots. For the first time tion is held fixed at 5% at the inner boundary and that their shown, Bz in the magnetosphere is primarily positive while relative outflow rate is relatively low, the O+ ions in the the IMF is negative. The first signs of the formation of a PSBL can at least temporarily reach relative densities as near-Earth neutral line are seen in Figure 10d with the high as nearly 50%. reconnection region appearing postmidnight. [46] When these ions are driven into the tail current sheet, [48] The reconnection region expands rapidly into the O+ rich streams are seen in the near-Earth region, as shown dawn and dusk sectors within 3 min. The dusk sector, in Figure 9c. The stream that forms near midnight is seen to though, is slower to fully reconnect, as evidenced by the convect around the duskside (Figure 9d), while at the same fact that the width of the reconnection regions necks down time the two closest cross-tail cuts in Figure 9d show a lobe in Figure 10e and the deepest regions of negative Bz at least

11 of 17 A09206 WINGLEE: ION CYCLOTRON AND HEAVY ION EFFECTS A09206

Figure 9. The density of O+ in the magnetosphere relative to the total ion number density. While the outflow rate of O+ is always smaller than that for H+, limitations in access of H+ to certain regions within the magnetosphere leads to regions that are substantially enhanced in O+, particularly during periods of strong southward IMF. These locally enhanced regions are also seen in the particle tracking results of Winglee [2003].

Figure 10. Evolution of Bz in the equatorial plane. A saturated color table is used to highlight the formation of the near-Earth neutral line. The black line shows the contour level for Bz = 0. The neutral line is seen to spread rapidly across the tail, with the dawnside initially becoming more negative than the duskside due to the fact that dawn-dusk convection tends to deplete the dawnside of plasma faster, which facilitates an enhanced reconnection rate.

12 of 17 A09206 WINGLEE: ION CYCLOTRON AND HEAVY ION EFFECTS A09206

Figure 11. Continuation of Figure 10. Plasmoid ejections occurs via a U-shaped neutral line, and within subsequent depolarization showing the presence of filamentary incursion of positive Bz down the tail, similar to the filaments seen in the plasma density. at early times are on the dawnside (Figure 10f). Note also rope is first seen to develop in the postmidnight sector. The that the reconnection does not occur in a homogeneous presence of the flux rope is due to the nonideal MHD terms manner but instead shows filamentary structures that track incorporated in the Ohm’s law and has been seen in other the density filaments described in the previous subsections. simulations [Zhu and Winglee, 1996; Winglee et al., 1998]. In other words the density structure and composition in the This ensuing plasmoid grows in size (Figure 12d) but is has tail is critical to how reconnection actually develops in the a curved front associated with the U-shaped reconnection region. region seen in Figure 11. Note also that the pressure [49] The inhomogeneities in the reconnection rate ends up contours show filamentary structure down the tail into the producing macroscopic changes in the ejection of the reconnection region. The filaments arise as plasma is plasmoid and associated structure in the tail. These global accelerated along the field lines to produce regions low in consequences are illustrated in Figure 11, which shows a both pressure and density. These filamentary structures continuation of the evolution of the reconnection region. It grow in width with the reconnection region. is seen that the development of the reconnection region [52] The flux rope or core magnetic field structure itself is expands tailward faster along the postmidnight sector, with best seen from the side, as shown in Figure 13. The the dawn sector following and the dusk sector being the structure basically consists of a flux rope through the center most tardy (Figures 11a–11c). This leads to the formation of the plasmoid with the By field being comparable to the Bz of U-shaped reconnection region, whereas as in ideal MHD within 3 Re of the core of the plasma (as evidence by the simulations the reconnection occurs almost uniformly pitch of the flux rope) and is less important further out. The across the tail. presence of this core magnetic field is consistent with [50] Partial depolarization occurs while the IMF is still particle simulations of Zhu and Winglee [1996] and has southward, as seen in Figures 11c–11e, with the expansion similar spatial variation of the core magnetic field reported of the region of positive Bz in the nightside inner magne- by Slavin et al. [1989] and Moldwin and Hughes [1991, tosphere from 15 Re to 22 Re. This recovery, like the 1992]. The plasmoid itself is relatively large with a full initial reconnection, is not homogeneous but instead has down-tail width of 20 Re. local regions of positive Bz with the duskside having a [53] After the plasmoid ejection, the reconnection activity longer tailward extension of positive Bz than the dawnside is dominated by the presence of small-scale flux ropes, as (Figures 11d and 11f). shown in Figure 14. These flux ropes appear 1 hour after [51] The effect of the above inhomogeneities on the the plasmoid is formed and ejected down the tail. The flux magnetic field structure is shown in Figure 12, which shows rope has a strong slant from the dawn to dusk sectors. This a top view of field lines superposed on contours of the total slant occurs because the dawn sector is depleted in density plasma pressure in the equatorial plane. From this top view, relative to the dusk sector and as such has a faster recon- reconnection is not seen until Figure 12c where a small flux nection rate than the dayside. The two frames show the

13 of 17 A09206 WINGLEE: ION CYCLOTRON AND HEAVY ION EFFECTS A09206

Figure 12. Evolution of the equatorial total plasma pressure and the mapping of the magnetic field in the tail. The U-shaped neutral is clearly manifested in the structure of the plasmoid, and behind it are filamented regions of very low plasma pressure. faster propagation of the portion of the flux rope on the using the electric and magnetic fields of global MHD dawnside relative to duskside. The slanting is also seen in simulation, despite the fact that the latter has symmetry contours of Bz in Figure 11f, where there is a sliver of plasma flows. positive Bz starting at 30 Re on the duskside and extending [55] In order to try and provide at least a partial resolution down the tail and crossing over the noon-midnight merid- of this dichotomy, we have investigated a multifluid treat- ian. These flux ropes have a much smaller scale size than ment of the global magnetosphere incorporating ion cyclo- the plasmoid with a lateral width of only a few Re, which is tron effects in both the generalized Ohm’s law and the ion comparable to the small flux ropes sizes reported by momentum equation. This present global solution is limited Lepping et al. [1996]. in that it has only a grid resolution in the inner magneto- sphere of 0.25 Re so that the ratio of grid spacing to the ion skin depth is of the order of 2–8 depending on the density 5. Discussion in the current sheet. Despite this limitation in grid resolu- [54] In this paper the possibility of whether small-scale tion, the above effects are sufficiently well resolved to ion cyclotron processes can produce macroscopic effects on produce dawn-dusk asymmetries in the plasma flows and the global magnetosphere has been quantitatively evaluated. auroral currents. This is a nontrivial problem because the typical ion gyro- [56] Key new results are as follows: radius of a keV proton is of the order of a few hundred [57] 1. Ion cyclotron and heavy ion effects produce a kilometers in the vicinity of a neutral sheet and even smaller distortion of the auroral currents for purely southward IMF in the stronger magnetized regions in the inner magneto- so that there is wrapping of the region 1 currents to the sphere. On the other hand the magnetosphere itself is tens to region 2 and 0 current systems across the noon-midnight a few hundred Re in size. Because of the disparity in scale meridian. This result differs from MHD in that the latter sizes ideal MHD is the main tool for global modeling, and case the currents are required to be antisymmetric across the in this methodology all ion cyclotron effects are neglected. noon-midnight meridian and as such the field-aligned cur- However, this assumption and the model results lead to an rents vanish along the noon-midnight meridian. The statis- interesting dichotomy. In particular for southward IMF one tical maps of the field-aligned auroral currents generated by attains a magnetosphere that has symmetric flows in the Iijima and Potemra [1978] indicate that the physically inner magnetosphere (neglecting corotation effects), which relevant result should have wrapping of the region 1 we know from observations is inaccurate. Single-particle currents. analysis of the current sheet acceleration also shows sub- [58] 2. The multifluid code shows dawn-dusk acceleration stantial dawn-dusk asymmetries [Speiser, 1965]. The dawn- of plasma in conjunction with the formation of the near- dusk acceleration is also seen in single-particle trajectories Earth neutral line. As mentioned above, MHD produces

14 of 17 A09206 WINGLEE: ION CYCLOTRON AND HEAVY ION EFFECTS A09206

Figure 13. Side view of the magnetic field mapping during plasmoid ejection, superposed on top the equatorial total plasma density. The presence of a strong core magnetic field in the y-direction leads to a flux rope structure in the center of the plasmoid, and this flux rope is curved around the U-shaped neutral point. symmetric flow, whereas as the single-particle analysis ropes, with the ion cyclotron corrections in the generalized shows that dawn-dusk acceleration of ions should be a Ohm’s law producing a strong core magnetic field. These driving force to the system. processes do not occur in ideal MHD. However, the core [59] 3. Because of the difference in gyroradius between magnetic field of the present simulations is seen in particle the heavy and light ionospheric ions, filamentary structures and hybrid simulations of reconnection and is also consist- down the tail are predicted, whereas ideal MHD simulations ent with the observations cited in the introduction. show a structureless reconnection. The presence of bursty [61] Thus the inclusion of finite ion cyclotron and heavy bulk flows suggests there is substantial structure occurring ion effects has global consequences that are consistent with in the magnetosphere. Further work needs to establish a link observations and which are omitted in ideal MHD treat- between model results and in situ observations, but the ments. It is important to note that the code indeed only present model makes a very important and testable predic- really resolves ion cyclotron effects near neutral regions. tion, specifically that there should be substantial enhance- This means that in the strong field regions of the inner ments in heavy ions densities in filamentary structures down magnetosphere, processes such as the formation of the the tail associated with reconnection occurring in the tail. symmetric ring current are not properly described. However, [60] 4. The reconnection does not occur in a uniform rate these strong field regions do not typically control the bulk across the tail due to the dawn-dusk acceleration. This acceleration of ions within the magnetosphere. Instead the nonuniformity in reconnection rate leads to skewed flux region that most influences the bulk particle acceleration

15 of 17 A09206 WINGLEE: ION CYCLOTRON AND HEAVY ION EFFECTS A09206

Figure 14. Continuation of Figure 13. A skewed flux rope with narrow width but long extension down the tail is seen after the plasmoid is ejected. and global flow (transport) is the reconnection region, and [64] Under the IMF conditions assumed here as well as in the physics of this region is at least partially resolved in the single-particle tracking results of Winglee [2003] the light present model. ions have problems accessing the tail reconnection region. [62] The question then reduces down to what is the Their fast velocity along the field lines tends to make them minimum scale length that is relevant to the reconnection convect beyond the near-Earth neutral line and therefore region, particularly in the tail, which has been the emphasis have limited access to the reconnection region. On the other in this paper. Clearly the most unstable current sheet in an hand the model results are consistent with the single-particle initial value problem is when the current sheet thickness is studies that show that heavy ions have excellent access and of the order of the ion skin depth which is beyond the locally can be the dominant ion species. This question of resolution of the global code. On the other hand the global light ion access to the reconnection region further empha- code is not considering an initial value problem, but instead sizes the importance of heavy ion mass loading in the it treats a driven system controlled by changes in the IMF. magnetotail. As such the system is driven from a thick current sheet to an [65] In summary, this paper demonstrates that ion cyclo- increasingly thinner current sheet until the point where tron terms in both the generalized Ohm’s law and the ion instabilities prevent further thinning. This inherent scale momentum equation are important to the dynamics of the length is presumably of the order of the ion skin depth. magnetosphere, especially in the presence of heavy iono- [63] However, the presence of heavy ions adds a new spheric ions. There are two areas where ion cyclotron terms potentially larger inherent scale length to the system. This have important consequences. The first is in the electric large scale length of the heavy ion gyroradius is resolved in field where they lead to the generation of the core magnetic the present multifluid simulations. Since these heavy ions field associated with the center of a plasmoid and within will become the first ions to become demagnetized, they flux ropes. The second is in the momentum equations where will be critical to the formation of the diffusion region the corrections lead to a preferential flow of plasma from the around the reconnection region. We have also demonstrated dawnside to the duskside as well as produce filamentary that even for minimal outflow rates, access to the recon- structures in the light ions. The physical manifestations of nection region is critical to understanding the reconnection these effects include a reconnection rate that varies across problem in the tail. the tail, a skewing of the flux ropes generated in the tail, and

16 of 17 A09206 WINGLEE: ION CYCLOTRON AND HEAVY ION EFFECTS A09206 the wrapping of the region 1 currents across the noon- Pritchett, P. L., F. V. Coroniti, and V. K. Decyk (1996), Three-dimensional midnight meridian into the region 2 or 0 currents. stability of thin quasi-neutral current sheets, J. Geophys. Res., 101, 27,413. Richtmyer, R. D., and K. W. Morton (1967), Difference Methods for Initial [66] Acknowledgments. This research was supported by NASA Value Problems, p. 300, Wiley-Interscience, Hoboken, N. J. grants NAG5-10962 and NAG5-11869 and NSF grant ATM-0105032 to Seki, K., T. Terasawa, M. Hirahara, and T. Mukai (1998), Quantification of the University of Washington. tailward cold O+ beams in the lobe/mantle regions with Geotail data: + [67] Lou-Chuang Lee thanks the two reviewers for their assistance in Constraints on polar O outflows, J. Geophys. Res., 203, 29,371. evaluating this paper. Seki, K., R. C. Elphic, M. Hirahara, T. Terasawa, and T. Mukai (2001), On atmospheric loss of oxygen ions from Earth through magnetospheric References processes, Science, 291, 5510. Shay, M. A., J. F. Drake, R. E. Denton, and D. Biskamp (1998), Structure of Birn, J., et al. (2001), Geospace Environmental Modeling (GEM) magnetic the dissipation region during magnetic reconnection, J. Geophys. Res., reconnection challenge, J. Geophys. Res., 106, 3715. 103, 9165. Biskamp, D., E. Schwartz, and J. F. Drake (1995), Ion-controlled collision- Sibeck, D. G. (1990), Evidence for flux ropes in the Earth’s magnetotail, in less magnetic reconnection, Phys. Rev. Lett., 75, 3850. Physics of Magnetic Flux Ropes, Geophys. Mongr. Ser., vol. 58, edited by Chappell, C. R., T. E. Moore, and J. H. Waite Jr. (1987), The ionosphere as C. T. Russell, E. R. Priest, and L. C. Lee, p. 637, AGU, Washington, D.C. a fully adequate source of the Earth’s magnetosphere, J. Geophys. Res., Sibeck, D. G., G. L. Siscoe, J. A. Slavin, E. J. Smith, S. J. Bame, and F. L. 92, 5896. Scarf (1984), Magnetotail flux ropes, Geophys. Res. Lett., 11, 1090. Cladis, J. B. (1986), Parallel acceleration and transport of ions from the Slavin, J. A., et al. (1989), CDAW-8 observations of plasmoid signatures in polar ionosphere to the plasma sheet, Geophys. Res. Lett., 13, 893. the geomagnetic tail: An assessment, J. Geophys. Res., 94, 15,153. Drake, J. F., R. G. Kleva, and M. E. Mandt (1994), Structure of thin current Slavin, J. A., C. J. Owen, M. Kuznetsova, and M. Hesse (1995), ISEE 3 layers: Implications for magnetic reconnection, Phys. Rev. Lett., 73, 1251. observations of plasmoids with flux rope magnetic topologies, Geophys. Elphic, R. C., C. A. Cattell, K. Takahashi, S. T. Bame, and C. T. Russell Res. Lett., 22, 2061. (1986), ISEE-1 and 2 observations of magnetic flux ropes in the magne- Sod, G. A. (1978), A survey of several finite difference methods for sys- totail: FTEs in the plasma sheet?, Geophys. Res. Lett., 13, 648. tems of nonlinear hyperbolic conservation laws, J. Comput. Phys., 27,1. Hones, J. W., Jr. (1976), The magnetotail: Its generation and dissipation, Speiser, T. W. (1965), Particle trajectories in model current sheets: 1. Ana- in Physics of Solar Planetary Environments, edited by D. J. Williams, lytical solutions, J. Geophys. Res., 70, 4219. p. 559, AGU, Washington, D.C. Steinolfson, R. S., and R. M. Winglee (1993), Energy storage and dissipa- Hones, J. W., Jr. (1979), Plasma flow in the magnetotail and their relation to tion in the magnetotail during substorms 2. MHD simulations, J. Geo- substorms theories, in Dynamics of the Magnetosphere, edited by S. I. phys. Res., 98, 7536. Akasofu, p. 545, D. Reidel, Norwell, Mass. Weimer, D. R., J. R. Kanm, and S.-I. Akasofu (1992), Variations of the Iijima, T., and T. A. Potemra (1976), The amplitude distribution of field- polar cap potential measured during magnetospheric substorms, J. Geo- aligned currents at northern high latitudes observed by Triad, J. Geophys. phys. Res., 97, 3945. Res., 81, 2165. Winglee, R. M. (2000), Mapping of ionospheric outflows into the magneto- Iijima, T., and T. A. Potemra (1978), Large-scale characteristics of field- sphere for varying IMF conditions, J. Atmos. Sol. Terr. Phys., 62, 527. aligned currents associated with substorms, J. Geophys. Res., 83, 599. Winglee, R. M. (2003), Circulation of ionospheric and solar wind particle Lepping, R. P., D. H. Fairfield, J. Jones, L. A. Frank, W. R. Paterson, populations during extended southward IMF, J. Geophys. Res., S. Kokubun, and T. Yamamoto (1995), Cross-tail magnetic flux ropes 108(A10), 1385, doi:10.1029/2002JA009819. as observed by the GEOTAIL spacecraft, Geophys. Res. Lett., 22, 1193. Winglee, R. M., and R. S. Steinolfson (1993), Energy storage and dissipa- Lepping, R. P., J. A. Slavin, M. Hesse, J. A. Jones, and A. Szabo (1996), tion in the magnetotail during substorms: 1. Particle simulations, J. Geo- Analysis of magnetotail flux ropes with strong core fields: ISEE 3 phys. Res., 98, 7519. observations, J. Geomagn. Geoelectr., 48, 589. Winglee, R. M., S. Kokubun, R. P. Lin, and R. P. Lepping (1998), Flux rope Lu, G., et al. (1996), High-latitude ionospheric electrodynamics as structure in the magnetotail: Comparison between Wind/Geotail observa- determined by the assimilative mapping of ionospheric electrodynamics tions and global simulations, J. Geophys. Res., 103, 135. procedure for the conjunctive SUNDIAL/ATLAS 1/GEM period of Winglee, R. M., D. Chua, M. Brittnacher, G. K. Parks, and G. Lu (2002), March 28–29, 1992, J. Geophys. Res., 101, 26,697. Global impact of ionospheric outflows on the dynamics of the magneto- Ma, Z. W., and A. Bhattacharjee (1996), Fast impulsive reconnection and sphere and cross-polar cap potential, J. Geophys. Res., 107(A9), 1237, current sheet intensification due to electron pressure gradients in semi- doi:10.1029/2001JA000214. collisional plasmas, Geophys. Res. Lett., 23, 1673. Yau, A. W., and M. Andre´ (1997), Source of ion outflow in the high latitude Moldwin, M. B., and W. J. Hughes (1991), Plasmoids as magnetic flux ionosphere, Space Sci. Rev., 80,1. ropes, J. Geophys. Res., 96, 14,051. Zhu, Z., and R. M. Winglee (1996), Tearing instability, flux ropes, and the Moldwin, M. B., and W. J. Hughes (1992), On the formation and evolution kinetic current sheet kink instability in the Earth’s magnetotail: A three- of plasmoids: A survey of ISEE 3 Geotail data, J. Geophys. Res., 97, dimensional perspective from particle simulations, J. Geophys. Res., 101, 19,259. 4885. Ogino, T., R. J. Walker, M. Ashour-Abdalla, and J. M. Dawson (1986), An MHD simulation pf the effects of the interplanetary magnetic field by component on the interaction of the solar wind with the Earth’s magneto- sphere during southward IMF, J. Geophys. Res., 91, 10,029. Øieroset, M. T. D. Phan, M. Fujimoto, R. P. Lin, and R. P. Lepping (2001), R. M. Winglee, Department of Earth and Space Sciences, University of In situ detection of collisionless reconnection in the Earth’s magnetotail, Washington, Seattle, WA 98195-1310, USA. ([email protected]. Nature, 412, 414. edu)

17 of 17